Hikvision Acusense Analytics Tested

By Rob Kilpatrick, Published Sep 23, 2019, 09:57am EDT

Hikvision touts "The Magic Behind It All" in their new Acusense line are 'deep learning algorithms' inside these cameras and recorders. But how well do these new models work?

free image1

We bought and tested the Hikvision DS-2CD2346G1-I/SL camera and DS-7608NXI-I2/8P/4S NVR for over three weeks, examining:

  • False alert performance: What did and did not cause false alerts?
  • Intrusion detection accuracy: Did it miss valid people or vehicles in the scene in multiple conditions?
  • Ease of setup: How difficult or time-consuming was setup and optimization?
  • Vs. competitive analytics: How did DeepinView compare to AI analytics such as Dahua Analytics+ and Wyze Cam and camera analytics from our IP Camera Analytics Shootout?
  • What components does it use? Ambarella, Huawei, Intel, NVIDIA, others?
  • How well does it integrate with major VMSes?

Readers should also see our tests of Hikvision's DeepinView AI analytics camera and DeepinMind NVR.

*******

**** *** ****** ** three ***** ** *******, Hikvision's ******** *** ******* positives:

  • ** ***** ****** ** Most ******* *******:******** ******* **** ** swaying ********/*******, ******* **** produced, *** **** ******** on *** **** ** the ****** *** *** trigger ***** ******.
  • ** ***** ****** ** Most ****** ****** *******:****** ****** ******* ** false ****** **** ** drastic ***** *******, ******* moving ******* *** *****, and ****** ******* ** and *** **** ******* throughout *******.
  • ******** ********* ** ****** Walking ******* ** ********:******** *** *** **** people ******* ******* ** crawling ******* *** ***** during *******, *** ** night.

*******, ** ******* **** significant ****** **** ***************** and ***** ******:

  • ****** ********** ** ********:****** ******* ******* *** scene **** ********* ********** as ********, **** ***** when ******** **** ** the **********. **** ***** result ** ****** ***** completely ****** ** ***** turn *** ******* ****** indoors, *** ********* ** detect ********.
  • ***** ****** ** *******:***** *** ***** ******* (ranging **** **** ******* deer) ******* ******* *** scene ********* ******** ***** alerts, ********** ** ******** used. ******* **** **** often ********** ** ********.
  • ***** ****** ** ****** Cars:****** **** ** *** scene **** ********* **** as ****** ******* **** passing ****** ****** ******* in ********.
  • ******** ***** ****** ** Sunlight ******* *******:******** ******* * ***** window *** **** ****** false ****** **** ** moving ******* ** *** scene.
  • ****** *** ******** ******* Through *****:****** ******* ******* *** scene **** ****** **********, day *** *****, ******* and ***.

Hikvision ********

**** **** ********* *** given *** *********** ** review ******* ****** ******* but *** *** ***** any *********** *** ******** changes ** ******* *********. They ******* *** **** Acusense ******* ****** **** in *********** **** ******** NVRs, *** ** *** tests ** *** ** difference ** *********** ******* Acusense *******' ** ***** detection *** ******** ******* used ** *********** **** the ***.

Settings ******* ****** ** *** ******

** *** ******, ** adjusted ******** ** *** if ********* *********** ********, but ***** **** ******* settings ******** *** **** results. ** *********'* ********* analytics, ********* ********, *** only ******** *********** *** threshold (**** ****** ** in ***** ****** **********) and *********** ("**** ** the ****** ***** *** trigger *****"). ******* ** these ******** ********** ***** alerts.

Vs. ***** *********+

*****'* *********+****** ************ ********* ********, with ** ***** ****** on ******* ** ****** which ********* *** ********* Acusense. ************* *** **** is *******, *********** **** requiring * **** ** be *** *** ****** and/or ******* ********* ********.

Vs. **** **

****'* **** ** ****** detection*********** ********, ******** *******, foliage, ******, *******, *** other *******. *******, **** has **** ***-******** *********** which **** ** ********** for ********** *** *** it ***** *********** ** third ***** ***.

Vs. ******** *** ***** ***

********'* ****-******** ***************** ***(*** ******* ** ***** ****** ********* ********) **** ************ ********, ignoring ****** **** ******* and ***** *** *** not ****** **** *** same ************** ****** ** Acusense.

Vs. ********* ******** ********* *** ***

******** ** *********'* ******** Intrusion *********, ******** *********** is ******* ******, ********* more ***** ****** **** more *******, ********* ******* and ******, **** ****** on *******, ***. *********** is *********** ****** **** standard ***, ***** ******* on *********** *** ***** change, ********* *** *** limited ** *** ****** false ***** ******* *****, without ************** ******* ******, vehicles, ** ***** *******.

*******

*** ********* **-*********-*/** *** be ***** ****** ***~$*** ******** *** **-*******-**/**/** (* channel/8 ***/* ******** *********) sells ***~$*** *** ******.

People ******** ******* *******/********, *** *******

******** ******** ****** ******* upright ********** ******* **** no ****** ****** *** detection ***** *** *** night:

******** ******** **** **** detected ** *** ***** using ******* ********.

Running ****** ******

*******, ****** ******* ******* the ***** **** ****** completely. ** ******** ******* improved **** *********.

Running-Subjects-Not-Detected

People ********** ************* ** ********

****** ******* ******* *** scene, ****** ** *** foreground/walking ****** *** ****** were ********** ********** ** vehicles. **** ***** **** to ********** ****** ********** if ***** ****** ** detect **** ******.

False ****** ** *******

******** ********** ******* ** both ***** *** ***** animals ** *** ***** of ****. ******* **** most ***** ********** ** vehicles, *** ********* ********** as ***** **** ******* toward *** ******.

False ****** ** ****** ****

****** *******, ******** ******* on ****** ******** **** cloud ******** ** ***** light ******* ********, **** though *** ******* *** not ****. ***** *********+ also ******** **** **** issue, ****** ** **** not **** ** ** other *********.

parked vehicle alert

False ****** ** ********

** *** *****, ******** alerted ** ******** ********* on *** ***** **** the ***** ****/*******. **** occurred *-* ***** *** day ** *** ****** of *** ****.

No ***** ****** **** *******, *******, ** ************

********, ******** ******* ***** alerts **** ****** ******* such ** ******* ******* and *** ******* **** produce, **** *****.

No-False-Alerts-On-Foliage_Shadows-Outdoors

******** **** ******* *** of *** **** ****** sources ** ***** ******, heavy **** *** **** dripping **** *** ****.

No-False-Alerts-On-Rain_Droplets

No ****** ***** ****** ** ****** ** *******

****** ******* ******** ******* all ****** ****** ***** alerts **** ** ******* light ******* ** *** scene, ***** *****.

******* **** ****** ****** through *** ***** **** also *******.

Detection ********

** *** *****, ******** reliably ******** ** ~***' distance, ~** *** ** the *** **-*********-*/** ****** we ******. ***** *** requirements *** ***, ********** to ***** ** ********, and ***** *************'* *** ********* *********, ***** ******** ****** double *** *** *** detection *** *** **** more ***** ***** ******.

Simple *****

***** ** *** **** step **** *********'* ****** intrusion ********* *****, **** requiring *** **** ** choose ** ****** ***/** vehicles ****** ** ******** in *** *****:

VMS ***********

******** ****** ****** ********* to*********'* ******* ********* *********** ***** ***** ***. These ****** **** ********* in ** ******* *** Milestone ********, *** *** analytics **** *** ********* in ***********, ******** ******* Center, ** ******* ******** Center.

Camera ********: ********* *** ******** ***** ****

*** ********* **-*********-*/** ****** uses ** ********* ***-*-**-** system ** ****, * common *** *** ***** cost ******.

Ambarella S3L System On Chip

*******, ** **** ******** a******** ************ ********** **** (***), unlike **** ***** **/******** models ***** **** ******** only * ****** ****. This ** ********' ******* generation, ******** ** **** (see *** ******:***** ******** ******* ***** Surveillance ******).

Versions ****

*** ********* ******** **** used ****** *******:

  • ********* **-*********-*/**: **.*.** ***** 181213
  • **-*******-**/**/**: **.*.** ***** ******

Comments (29)

It looks like there is a way to turn off the detection for vehicle or human so to be fair I doubt many would be looking for autos in a small hallway scene. However, that error leads me to believe they may be using the size of an object in the scene to differentiate between human or auto?

Agree
Disagree
Informative: 1
Unhelpful
Funny

I definitely want to know if there is a car in my mall!

Agree: 1
Disagree: 1
Informative
Unhelpful
Funny: 5

For sure, if I set up a camera in that hallway in production, I would likely not have checked the "vehicle" detection box. That's exactly the problem. If we had checked only person detection in that scene, we would have missed a significant number of subjects. They simply aren't detected at all.

Agree: 1
Disagree
Informative: 3
Unhelpful
Funny

It also leads me to believe that they are not doing any kind of true object classification, most commonly done today with CNN/DNN analytics. Kind of curious given their use of the Movidius chip, from this test they do not seem to be fully utilizing its capabilities.

Agree: 1
Disagree
Informative: 3
Unhelpful
Funny

We've added a section comparing Acusense to Hikvision's standard Intrusion and VMD:

Vs. Hikvision Standard Intrusion And VMD

Compared to Hikvision's standard Intrusion analytics, Acusense performance is clearly better, rejecting more false alarms from more sources, including shadows and lights, more alarms on animals, etc. Performance is drastically better than standard VMD, which alerted on practically any pixel change, including but not limited to the common false alert sources above, without discriminating between people, vehicles, or other objects.

Agree
Disagree
Informative: 3
Unhelpful
Funny

I can confirm the drastic improvement in analytics and overall picture quality compared to the previous generation. Good review, thank you!

I hope they will improve the detection algorithms in the future FW versions

Agree
Disagree
Informative
Unhelpful
Funny

Interestingly enough, the Hik website I primarily use does not even show a 5.5.72 firmware. It shows the current one now as 5.6.3, and the only previous version shown is 5.6.0. Ditto for the other Accusense models I checked. (5.6.3 is also available for many other cameras without Accusense.)

4 MP Outdoor AcuSense Fixed Turret Camera | Hikvision US | The world’s largest video surveillance manufacturer

Agree
Disagree
Informative
Unhelpful
Funny

Animals were sometimes classified as human...However, people running through the scene were missed completely. No settings changes improved this detection.

Guard 1: Busy shift?

Guard 2: Had an alert, turned out to be a dog. Otherwise, nothing...

Agree
Disagree
Informative
Unhelpful
Funny: 4

Hey Ethan. Not sure if you want to try this, but do you know what would happen if you crank sensitivity up to 90, or 100? Just curious if that might help the camera catch fast moving objects.

Agree
Disagree
Informative
Unhelpful
Funny

When I tested this I did adjust sensitivity and threshold to try and get an alarm while I was running, even having it at max wouldn't trigger an alarm, it did not even draw a bounding box around me.

Agree
Disagree
Informative: 1
Unhelpful
Funny: 1

...get an alarm while I was running, even having it at max wouldn't trigger an alarm, it did not even draw a bounding box around me.

SoC: Did you see that?

AI: See what?

SoC: If I knew “what”, I wouldn’t need you. Something just went flying by...

AI: You mean that blurred streak of light. I have no idea. It’s gone now. Anyway, out of frame, out of mind, right?

SoC: Well I’m gonna send an alert anyway, the user should be notified!

AI: And what are you going to say it is this time, a giraffe, a car, maybe a sand storm?

SoC: Nah, you’re right those are getting stale. You got any ideas?

AI: Yeah, keep your trap shut. Remember, if faced with the choice of false negative vs. false positive, ALWAYS choose false negative. If you missed it, they probably will too; nobody’s the wiser. On the other hand, calling out “heavy snow” in July will just put us on the cover of IPVM, again!

SoC: You’re so smart!

Agree
Disagree
Informative
Unhelpful
Funny: 6

Thanks for the information Rob.

Agree
Disagree
Informative
Unhelpful
Funny

Hi guys - One of our common false triggers is insects at night. If you could figure out how to test for this it would be very helpful.

Agree: 6
Disagree
Informative
Unhelpful
Funny

Thanks, IPVM for testing different lines.

There is one thing I do not understand.
Analytics got settings, right?
And those probably need to be tuned for different conditions.
Are you sure you do a good job on tunning it?
Or do you always use out of the box ones?

Agree: 1
Disagree
Informative
Unhelpful
Funny

Are you sure you do a good job on tunning it?

We spoke repeatedly with Hikvision management throughout this test and incorporated their feedback and suggestions. Ethan will update the report with details.

Agree
Disagree
Informative: 1
Unhelpful
Funny

Thanks. I assume you do the same with other manufacturers, so you do adjust settings form scene to scene.

Agree
Disagree
Informative
Unhelpful
Funny

For clarity, we've added two sections to the report. As in other tests, we reached out to Hikvision for comment and suggestions on optimizing the analytics. We also perform our own settings changes to see what effects they have (in this case, minimal to none).

Hikvision Feedback

Note that Hikvision was given the opportunity to review results during testing but did not offer any suggestions for settings changes to improve detection. They pointed out that Acusense cameras worked best in conjunction with Acusense NVRs, but in our tests we saw no difference in performance between Acusense cameras' on board detection and Acusense cameras used in conjunction with the NVR.

Settings Changes Tested In All Scenes

In all scenes, we adjusted settings to see if detection performance improved, but found that default settings produced the best results. In Hikvision's intrusion analytics, including Acusense, the only settings adjustments are threshold (time object is in scene before triggering) and sensitivity ("size of the object which can trigger alarm"). Neither of these settings eliminated false alerts.

Agree: 1
Disagree
Informative: 5
Unhelpful
Funny

Hi Ethan,

My understanding of Hik's Acusense is you can get these analytics on-board the camera as well as built into the NVR. The NVR will work the analytics on ordinary camera feeds, while Acusense cameras will provide the analytics on NON-Acusense NVRs by just pushing through an event notification.

Is this correct? If so I'm thinking of picking up one of the Acusense NVRs to replace the one we have at the shop which has various IP cameras with the basic analytics. Thoughts?

Agree
Disagree
Informative
Unhelpful
Funny

Yes, UI3, this is correct, the Acusense NVR analytics will work with third party camera streams and provide people and vehicle alerts. The Acusense cameras will have event notifications on non-Acusense NVRs but won't provide the classification.

Agree
Disagree
Informative: 2
Unhelpful
Funny

Thanks very much for the clarification!

Agree
Disagree
Informative
Unhelpful
Funny

Hello,

I believe there might be a typo in the "Vs. Wyze AI" section of the article. You mentioned "Axis Perimeter Defender" should that be Acusense instead?

Thanks,

Agree
Disagree
Informative: 1
Unhelpful
Funny

Yes, you are right. I've fixed that, thanks for pointing it out!

Agree
Disagree
Informative
Unhelpful
Funny

You said :

Detection Distance

In our tests, Acusense reliably detected at ~105' distance, ~10 PPF in the 4MP DS-2CD2346G1-I/SL camera we tested

Question:

Can please specify the lens used in the test? 2.8mm, 4mm or 8mm?

Agree
Disagree
Informative
Unhelpful
Funny

10 ppf at 100' - it's 2.8mm

Agree
Disagree
Informative
Unhelpful
Funny

This is correct, we used a 2.8mm for the test.

Agree
Disagree
Informative
Unhelpful
Funny

Any updates using the later 5.6.0 or 5.6.3 firmware? This will indeed be a huge PR blow for Hik if it cannot match Dahua Analytics+.

The Dahua A+ models I see also include the larger sensor (1/1.8), where Hik is using the smaller size similar to their mid to lower end models.

Agree: 1
Disagree
Informative
Unhelpful
Funny

Is it still the case that standard motion events (VMD) is disabled when the AcuSense analytics are enabled resulting in continuous recording? I tested the DS-2CD2T46G1 when it first came out and could not have both VMD and AcuSense enabled simultaneously.

Agree
Disagree
Informative: 1
Unhelpful
Funny

UI5, I updated firmware on our DS-2CD2346G1-I/SL to V5.6.0 build 191111 and I don't see an option for VMD, even when I turn off smart events in the metadata settings.

Agree
Disagree
Informative: 1
Unhelpful
Funny

That's too bad. It's not a huge deal if you have lots of HDD space using an NVR, but the continuous recordings using a MicroSD fills up quickly.

Agree
Disagree
Informative
Unhelpful
Funny
Read this IPVM report for free.

This article is part of IPVM's 7,338 reports and 973 tests and is only available to subscribers. To get a one-time preview of our work, enter your work email to access the full article.

Already a subscriber? Login here | Join now
Loading Related Reports